Tracking of multiple maneuvering targets in a dense environment

Abstract

This paper discusses a multitarget tracking algorithm for use in a dense environment, where there exist other targets or clutter near the tracking target. It is currently one of the most important technical problems in the air traffic control. The joint probabilistic data association (JPDA) algorithm, considered to be a promising multitarget tracking algorithm, has the deficiency that the tracking of the multiple maneuvering targets in a dense environment is difficult. This paper proposes an algorithm in which the multiple motion models are obtained by using different constant acceleration vectors under the JPDA approach; six-dimensional motion models composed of the position and the velocity of the target in three-dimensional space are used in parallel. In this algorithm, the fuzziness due to the nonuniqueness of the motion model when the target maneuvers can be expressed in the calculation of the gain matrix, the covariance matrix and the motion states of the target. In addition, it is possible to adjust the prediction region for the observed vector (center and spread), according to the maneuvering of each target. Simulations indicate a remarkable improvement in the rate of tracking success by the proposed algorithm, reaching several times to ten times larger than that obtained in JPDA. © 1997 Scripta Technica, Inc. Electron Comm Jpn Pt 1, 80 (5): 23–34, 1996